|Prior work on Branch Profiling/Optimization email@example.com (1992-07-10)|
|Re: Prior work on Branch Profiling/Optimization firstname.lastname@example.org (1992-07-13)|
|From:||email@example.com (Kjell Post)|
|Organization:||University of California, Santa Cruz (CE/CIS Boards)|
|Date:||Fri, 10 Jul 1992 21:23:28 GMT|
I'd like to hear from people who have references to or experience with
the following optimization technique, called "branch profiling":
If a compiler knows which branch is executed more often
in a condition, it can generate better code.
This can be arranged by having the program emit profile
data during a couple of sample runs, and then recompile it
with branching statistics available to the compiler.
In the compilers archive I've seen references to a thesis by Karr,
a program called PTOOL from IBM, and a compiler from National
Semiconductor. Again, if you're familiar with this technique,
and know when and where it first appeared, I would really appreciate
hearing from you!
Return to the
Search the comp.compilers archives again.